960 volt, 1MHz sine wave?

[marfire]

I'm trying to determine what the best way to produce a 960 volt, 1MHz sine wave would be. The wave needs to be balanced so it's 480 v above ground and 480 v below ground. The current requirements are minimal/negligible. I just need the voltage. So far I've tried winding a transformer (wound 1:40) with a split primary, where I've put 24 volts in. The transistors on each primary are being triggered by a crystal driving a 4017, so the transistors are firing on 0 and 2 respectively, on the 4017's sequence. I've been having problems with inductance losses in my transformer at this frequency.

My question is whether there's another way to produce this that's more efficient? I was thinking perhaps two inductors, one firing positive and the other negative, but I can't find any literature about how to make the spark from an inductor that precise. Any help is appreciated!

 

[alec_t]

I'm curious as to what this whopping wave will be used for?

 

[BeerBelly]

A sine wave osc. is more efficient than a square wave and a large toroid is very efficient.
1 mhz is right in the middle of the AM broadcast band and will interfere with radio reception,

 

[marfire]

It's for charging a capacitor plate in something similar to a Cockroft-Walton generator.

 

[marfire]

Thanks BeerBelly. This will be operating in a shielded room so hopefully there won't be too much RF interference. I'll try your suggestion of the sine wave oscillator this week.

 

[kubeek]

"The current requirements are minimal/negligible." and "It's for charging a capacitor plate in something similar to a Cockroft-Walton generator." are completely opposite statements, especially at this sort of voltage and frequency. What EXACTLY is the load going to be?
I doubt you will be able to find a diode that can withstand 1kV and work at 1MHZ to make that multiplier. So you want to somehow charge a capacitor to 1kV? Or something else? Why the 1MHz requirement?

 

[marfire]

Well, it's not a Cockroft-Walton generator. It was probably a poor choice of analogy, since I'm not using it for a voltage multiplier. I just had in mind the function of the capacitor in a CW which pushes and pulls a current. I just have a pair of small aluminum plates coated in titanium dioxide (in a shielded box) where one is connected to ground and the other is connected to that voltage. It will be used to test the dipole constant of a product (some chemicals) and their reaction to radio frequency, and to see if there will be any problems with induced current in the product. The capacitance between them is adjustable by altering the distance between the plates.

 

[kubeek]

What is the range of the capacitance? Even just 100pF has impedance 1600ohm @ 1Mhz, so there will be around 300mA flowing through it with that voltage.

 

[marfire]

Approx 8pF to 83pF. Even if I needed 1000mA and 960 watts I'd like to be able to know how do that best, whether it be a large toroid, two inductors, etc. I'm a biochemist and not an EE so I don't have all the tricks up my sleeve regarding this stuff like you guys have.

 

[marfire]

Actually I got that capacitance wrong. It's between 83pF and approximately 183pF. I had to double check my notes. Obviously the capacitance will change depending upon the chemical between the plates as well.

 

[Dean Huster]

960v @1000ma? Sounds like a kilowatt linear amplifier to me!

 

[ronsimpson]

You can't get silicon that will work that fast at that high a voltage. Your transformer idea is what I would do.

 

[marfire]

Thanks for the helpful responses! For the time being we're going to try a circuit similar to a Solid State Tesla Coil. Well, really it is a SSTC, we're just aiming for lower voltages than TC's usually produce so we'll have a much smaller turns ratio on the second TF than a TC usually has. We're going to use the transformer we currently have, at a vastly lower frequency, rectified to DC, and then use LC resonance to get the frequency we need, with a manually adjustable piece of ferrite to control inductance and frequency. We'll see how it goes. Thanks again for the help, everyone!

 

[JimB]

Coming a little late into this discussion, can I suggest something which should be quite easy, if you can find the parts.

Have a look at the attachment.

The sample under test is connected across a parallel tuned circuit which is resonant at the frequency of interest (1 Mhz ?).
I have shown the capacitor as a "split stator" type, so that the adjustment control can be at earth potential.

The oscillator can be whatever you have available, the amplifier I was thinking of using a valve (tube for those across the Atlantic).
The valve amplifier will work with high voltages without any problems, and link coupling the amplifier to the parallel tuned circuit will give you plenty of volts across the sample under test.

JimB

 

[marfire]

JimB, this is brilliant! It's exactly the kind of thing I was looking for! I have to admit the tube/valve thing is a tad daunting to me, not being an EE, but really this seems like precisely what I need. Maybe if I do some homework and figure out how to use valves beyond mere theory I can make that guitar amp I've always wanted to make. Thank you!

 

[dr pepper]

Could you use something like a resonant spark gap transmitter?, should be reasonably simple to contruct, no good if you want a linear amp though.
You can get silicon that will work at those freq's, but its highly specialized, difficult to obtain and costs, there are some butterfly fets that will work up to uhf freqs at high voltages.
Transmitting triodes are probably also difficult to get, and require carefull design.

 

[marfire]

I had looked into using a spark gap, but like you said it doesn't work for this. I managed to source some fet modules that will handle the amplifier stage. They were pretty expensive but as this is only a one-off research apparatus and not a production piece, it was tolerable. I did look into using tubes for the amp, and they're vastly less complicated than I had thought, but the fets were still easier to implement.

 

[dr pepper]

Your obviously not a noob if you sourced some hvhf fets.
Dont be fooled about the simplicity of a valve output stage, theres still a lot of pitfalls in the design after the calculations are done, high power thermionics is a bit of an art, but I spose you could say that of anything high power.
I'm sure we'd all like to hear your progress, and watch out for rf burns.